Talk:Retrieval-based Voice Conversion
Add topicLatest comment: 1 year ago by 2003:DA:CF2E:4539:F0CC:8548:5B09:43DD in topic Significant re-write required
 | This article is rated Start-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects: | |||||||||||||
| ||||||||||||||
Significant re-write required
[edit]This article requires a significant re-write:
- First off, while RVC is *ALSO* a specific algorithm on how to train a voice model, it's mainly a special *FORMAT* of voice model, akin to how WAV is a special *FORMAT* of audio file.
- Second, this article links not to the original RVC 1.0 or 2.0, but to one of a number of simple GUI frontend forks making use of RVC, and uses this fork's release date as the date of release for RVC. The original RVC 1.0 and 2.0, without a GUI, were released in February and Autumn of 2023, respectively.
- Third, this article acts like RVC, and specifically this GUI frontend, would be the only format or software for voice conversion, when in fact voice.ai (with their own proprietary format) opened their final beta version for the public in early-October 2022, as the very first publically available service for voice conversion which didn't require a subscription costing several hundred or even several thousand Dollars a month (depending upon your company's taxable revenue, as was the case with commercial services such as ReSpeecher).
- Fourth, there are also other frontends to use the RVC format, namely Replay.
- Finally, the entire article fails to clearly describe why we should care, i. e. the two decisive advantages the RVC format has compared to competitor formats: a.) You need far less source material for a robust voice, and b.) multi-language capability is already built in, so you can use voices in the RVC format for more languages than English, Mandarine, or Korean only. --2003:DA:CF2E:4539:F0CC:8548:5B09:43DD (talk) 03:53, 21 October 2024 (UTC)